Classification and Skimming of Articles for an Effective News Browsing

摘要

In order to browse the news video effectively, classification and skimming of news articles are positively essential. In this paper, we propose the classification and skimming of articles for an effective news browsing. The classification method uses tags to distinguish speakers in the closed-caption. The skimming method extracts the representative sentence from the part of article introduced by the anchor in the closed-caption and the representative frames consisting of anchor frame, open-caption frames, and frames synchronized with high-frequency terms. In the experiment, we have applied the proposed classification and skimming methods to news video with Korean closed-captions, and have empirically confirmed that the proposed methods could support effective browsing of news videos.